AUTOMATED YIELD PREDICTION USING LSTMS AND REMOTE SENSING DATA

This research presents an automated approach for predicting crop yield using LSTM networks and remote sensing data. The study leverages the temporal dynamics of satellite imagery, weather data, and soil properties to develop a robust predictive model capable of capturing complex dependencies in agricultural systems. The LSTM framework is utilized to model the sequential patterns present in time-series data, enabling accurate yield forecasts by learning from historical trends and environmental factors. The dataset comprises historical remote sensing data, including spectral indices like NDVI, along with meteorological and soil data, which are preprocessed and normalized to optimize model performance and enhance predictive accuracy. The model is trained and validated using an 80-20 data split, ensuring a balanced evaluation of its generalization capability. Performance is tested using evaluation parameters such as MAE, RMSE, R², and MAPE, providing a comprehensive analysis of predictive reliability. The results demonstrate that LSTM networks significantly outperform traditional statistical methods in crop yield forecasting, showcasing their potential in handling complex, multi-source agricultural data. This work contributes to precision agriculture by providing a scalable and efficient tool for yield estimation, aiding in resource management, strategic planning, and data-driven decision-making to support sustainable farming practices